INDEX

Explanations

phrases indicating a directive or action

oai_token-act-pair · gpt-3.5-turbo

repetitive phrases emphasizing the word "just."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ixel

-0.70

lav

-0.67

untled

-0.67

ensical

-0.66

 plaintiff

-0.63

endra

-0.62

glomer

-0.62

 adversary

-0.60

pora

-0.60

 PLUS

-0.59

POSITIVE LOGITS

ifiable

1.05

ifications

1.03

 kidding

0.91

if

0.89

ifi

0.85

ices

0.85

 plain

0.81

icia

0.77

ific

0.76

itia

0.74

Activations Density 0.098%

phrases indicating a directive or action

repetitive phrases emphasizing the word "just."

No Comments

No Known Activations